2021-09-09 02:38:46

by John Stultz

Subject: [PATCH] dma-buf: system_heap: Avoid warning on mid-order allocations

When trying to do mid-order allocations, set __GFP_NOWARN to
avoid warning messages if the allocation fails, as we will
still fall back to single page allocatitions in that case.
This is the similar to what we already do for large order
allocations.

Cc: Daniel Vetter <[email protected]>
Cc: Christian Koenig <[email protected]>
Cc: Sumit Semwal <[email protected]>
Cc: Liam Mark <[email protected]>
Cc: Chris Goldsworthy <[email protected]>
Cc: Laura Abbott <[email protected]>
Cc: Brian Starkey <[email protected]>
Cc: Hridya Valsaraju <[email protected]>
Cc: Suren Baghdasaryan <[email protected]>
Cc: Sandeep Patil <[email protected]>
Cc: Daniel Mentz <[email protected]>
Cc: Ørjan Eide <[email protected]>
Cc: Robin Murphy <[email protected]>
Cc: Simon Ser <[email protected]>
Cc: James Jones <[email protected]>
Cc: Leo Yan <[email protected]>
Cc: [email protected]
Cc: [email protected]
Signed-off-by: John Stultz <[email protected]>
---
drivers/dma-buf/heaps/system_heap.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/dma-buf/heaps/system_heap.c b/drivers/dma-buf/heaps/system_heap.c
index 23a7e74ef966..f57a39ddd063 100644
--- a/drivers/dma-buf/heaps/system_heap.c
+++ b/drivers/dma-buf/heaps/system_heap.c
@@ -40,11 +40,12 @@ struct dma_heap_attachment {
bool mapped;
};

+#define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO | __GFP_COMP)
+#define MID_ORDER_GFP (LOW_ORDER_GFP | __GFP_NOWARN)
#define HIGH_ORDER_GFP (((GFP_HIGHUSER | __GFP_ZERO | __GFP_NOWARN \
| __GFP_NORETRY) & ~__GFP_RECLAIM) \
| __GFP_COMP)
-#define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO | __GFP_COMP)
-static gfp_t order_flags[] = {HIGH_ORDER_GFP, LOW_ORDER_GFP, LOW_ORDER_GFP};
+static gfp_t order_flags[] = {HIGH_ORDER_GFP, MID_ORDER_GFP, LOW_ORDER_GFP};
/*
* The selection of the orders used for allocation (1MB, 64K, 4K) is designed
* to match with the sizes often found in IOMMUs. Using order 4 pages instead
--
2.25.1

2021-09-14 13:58:01

by Daniel Vetter

Subject: Re: [PATCH] dma-buf: system_heap: Avoid warning on mid-order allocations

On Thu, Sep 09, 2021 at 02:37:41AM +0000, John Stultz wrote:
> When trying to do mid-order allocations, set __GFP_NOWARN to
> avoid warning messages if the allocation fails, as we will
> still fall back to single page allocatitions in that case.
> This is the similar to what we already do for large order
> allocations.
>
> Cc: Daniel Vetter <[email protected]>
> Cc: Christian Koenig <[email protected]>
> Cc: Sumit Semwal <[email protected]>
> Cc: Liam Mark <[email protected]>
> Cc: Chris Goldsworthy <[email protected]>
> Cc: Laura Abbott <[email protected]>
> Cc: Brian Starkey <[email protected]>
> Cc: Hridya Valsaraju <[email protected]>
> Cc: Suren Baghdasaryan <[email protected]>
> Cc: Sandeep Patil <[email protected]>
> Cc: Daniel Mentz <[email protected]>
> Cc: ?rjan Eide <[email protected]>
> Cc: Robin Murphy <[email protected]>
> Cc: Simon Ser <[email protected]>
> Cc: James Jones <[email protected]>
> Cc: Leo Yan <[email protected]>
> Cc: [email protected]
> Cc: [email protected]
> Signed-off-by: John Stultz <[email protected]>

Acked-by: Daniel Vetter <[email protected]>

> ---
> drivers/dma-buf/heaps/system_heap.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/dma-buf/heaps/system_heap.c b/drivers/dma-buf/heaps/system_heap.c
> index 23a7e74ef966..f57a39ddd063 100644
> --- a/drivers/dma-buf/heaps/system_heap.c
> +++ b/drivers/dma-buf/heaps/system_heap.c
> @@ -40,11 +40,12 @@ struct dma_heap_attachment {
> bool mapped;
> };
>
> +#define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO | __GFP_COMP)
> +#define MID_ORDER_GFP (LOW_ORDER_GFP | __GFP_NOWARN)
> #define HIGH_ORDER_GFP (((GFP_HIGHUSER | __GFP_ZERO | __GFP_NOWARN \
> | __GFP_NORETRY) & ~__GFP_RECLAIM) \
> | __GFP_COMP)
> -#define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO | __GFP_COMP)
> -static gfp_t order_flags[] = {HIGH_ORDER_GFP, LOW_ORDER_GFP, LOW_ORDER_GFP};
> +static gfp_t order_flags[] = {HIGH_ORDER_GFP, MID_ORDER_GFP, LOW_ORDER_GFP};
> /*
> * The selection of the orders used for allocation (1MB, 64K, 4K) is designed
> * to match with the sizes often found in IOMMUs. Using order 4 pages instead
> --
> 2.25.1
>

--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

2021-09-14 14:59:27

by Sumit Semwal

Subject: Re: [PATCH] dma-buf: system_heap: Avoid warning on mid-order allocations

Thanks John!

On Tue, 14 Sept 2021 at 19:26, Daniel Vetter <[email protected]> wrote:
>
> On Thu, Sep 09, 2021 at 02:37:41AM +0000, John Stultz wrote:
> > When trying to do mid-order allocations, set __GFP_NOWARN to
> > avoid warning messages if the allocation fails, as we will
> > still fall back to single page allocatitions in that case.
> > This is the similar to what we already do for large order
> > allocations.
> >
> > Cc: Daniel Vetter <[email protected]>
> > Cc: Christian Koenig <[email protected]>
> > Cc: Sumit Semwal <[email protected]>
> > Cc: Liam Mark <[email protected]>
> > Cc: Chris Goldsworthy <[email protected]>
> > Cc: Laura Abbott <[email protected]>
> > Cc: Brian Starkey <[email protected]>
> > Cc: Hridya Valsaraju <[email protected]>
> > Cc: Suren Baghdasaryan <[email protected]>
> > Cc: Sandeep Patil <[email protected]>
> > Cc: Daniel Mentz <[email protected]>
> > Cc: Ørjan Eide <[email protected]>
> > Cc: Robin Murphy <[email protected]>
> > Cc: Simon Ser <[email protected]>
> > Cc: James Jones <[email protected]>
> > Cc: Leo Yan <[email protected]>
> > Cc: [email protected]
> > Cc: [email protected]
> > Signed-off-by: John Stultz <[email protected]>
>
> Acked-by: Daniel Vetter <[email protected]>
>
Pushed to drm-misc-next.

> > ---
> > drivers/dma-buf/heaps/system_heap.c | 5 +++--
> > 1 file changed, 3 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/dma-buf/heaps/system_heap.c b/drivers/dma-buf/heaps/system_heap.c
> > index 23a7e74ef966..f57a39ddd063 100644
> > --- a/drivers/dma-buf/heaps/system_heap.c
> > +++ b/drivers/dma-buf/heaps/system_heap.c
> > @@ -40,11 +40,12 @@ struct dma_heap_attachment {
> > bool mapped;
> > };
> >
> > +#define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO | __GFP_COMP)
> > +#define MID_ORDER_GFP (LOW_ORDER_GFP | __GFP_NOWARN)
> > #define HIGH_ORDER_GFP (((GFP_HIGHUSER | __GFP_ZERO | __GFP_NOWARN \
> > | __GFP_NORETRY) & ~__GFP_RECLAIM) \
> > | __GFP_COMP)
> > -#define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO | __GFP_COMP)
> > -static gfp_t order_flags[] = {HIGH_ORDER_GFP, LOW_ORDER_GFP, LOW_ORDER_GFP};
> > +static gfp_t order_flags[] = {HIGH_ORDER_GFP, MID_ORDER_GFP, LOW_ORDER_GFP};
> > /*
> > * The selection of the orders used for allocation (1MB, 64K, 4K) is designed
> > * to match with the sizes often found in IOMMUs. Using order 4 pages instead
> > --
> > 2.25.1
> >
>
> --
> Daniel Vetter
> Software Engineer, Intel Corporation
> http://blog.ffwll.ch


Best,
Sumit.